Page 1 of 1

[SOLVED]use of bounds in anamorphosis

PostPosted: Tue Aug 20, 2013 7:37 am
by MatthewCobb
Hi Didier / Fabian et al.,

I seem to be firing out a lot of questions lately - so thankyou for your help!

My next query relates to use of bounds in anamorphosis. By setting flag.bound=TRUE, does this coerce / force z2y and y2z to yield transformation values within defined azmax/azmin and aymax/aymin?

The issue I am running up against is that the back transformed values from a series of gaussian simulations are yielding significant percentages of z values that are well outside the realms of reality.
I have been trying to constrain the simnulation results to realistic values by constraining both azmax/min and pzmax/min to the respective max and min of the input raw data during the original z2y transform, however it appears that the resulting point anamorphosis model does not adhere to the limits I have specified.

Have I either misinterpretetd what flag.bound does, or am I not specifying the limits to adhere to during transformation correctly?

As always, help is much appreciated.

Thanks,

Matt.

Re: use of bounds in anamorphosis

PostPosted: Tue Aug 20, 2013 8:54 pm
by Didier Renard
Hello

When fitting the anamorphosis based on a set of data (using anam.fit function), the uer can provide some bounds which superseed the ones derived from the data themselves. This is through the parameters azmin and azmax (this is on the side of the variable values) and aymin and aymax (this is for the frequencies).
Then the user can define azmin (which must be smaller than the minimum value of the data values). Similarly, he can provide azmax (larger than the upper bound of the data). Finally he can provide aymin which must be smaller than G(1/N) where N stands for the number of data.

This bounds (when defined) are stored in the R object containing the anamorphosis.

Then, when using this R-object in the methods anam.z2y, anm.y2z or anam.eval, the user can specify if the bounds (aymin, aymax, azmin and azmax) are to be used or not.
If one of these bounds (say azmin) must be used (flag.bound=TRUE) and if this bound has been explicitely defined, then the result is compared with this bound: i.e. if the current procedure is anam.y2z() and if the result of the conversion from the y input value, using the Hermite polynomials, ends up to a value larger than azmax, then the value azmax is returned.

Hoping this information is clear enough.